Causal effects of coupon expiration on consumer acceptance: evidence from heterogeneous treatment effects in an observational scenario survey

This paper examines how coupon expiration time affects consumer acceptance using observational scenario-survey data and causal machine learning. Because the mechanism assigning the expiration attribute is undocumented and responses are stated acceptance in hypothetical scenarios, we treat the two-hour (versus one-day) window as an observational exposure identified under selection-on-observables, supported by cross-fitted propensity scores, balance diagnostics, and sensitivity analysis. The Context-Contingent Time Preference (CCTP) framework supplies directional predictions about situational moderation. Inverse-probability weighting estimates the two-hour window reduces stated acceptance by 18.3% points (cluster-robust SE =0.014); propensity score matching yields ATT =−0.160 (5,305 pairs). On a held-out confirmation sample, the penalty is pervasive but amplified by urgency (βˆ=−0.087, SE =0.041) and destinations over 25 minutes away (θˆ=−0.147, SE =0.041), most severe at the intersection of solitude and urgency (−0.134, SE =0.032), and attenuated by social presence (+0.088, SE =0.022; composition-adjusted +0.047, SE =0.032). It concentrates in restaurant and coffee-house coupons and is absent for bar coupons. Simulations show effect-based targeting reduces conversion losses (29–42%) versus random allocation; propensity-based targeting performs worse than random. An E-value of 2.13 benchmarks sensitivity. Results concern stated, not revealed, acceptance.